Voice AI agents with ElevenLabs that hold up in production.
I build AI phone assistants and voice agents on the ElevenLabs platform, review existing agents for quality and risk, and advise you on how your use case fits the EU AI Act.
- AgentHello, you are speaking with the city utility’s digital assistant. How can I help?
- CallerI’d like to submit my meter reading.
- AgentSure. For security, what is your customer number?
- Identity verified, meter reading being submitted
Three ways I can help
From the first prototype to an agent that handles thousands of calls a day. You decide whether I build, review or advise.
-
Voice AI agent development
I design and implement your agent on the ElevenLabs platform, with a clear conversation flow, connected systems and testing before real customers call.
- Conversation design, prompt and voice
- Webhook tools for CRM, calendar and ticketing
- Identity verification before sensitive actions
- Telephony, web widget and human handover
-
Review, analysis and consulting
Your agent is live but not performing as it should? I analyse conversations, configuration and tool logic and tell you exactly what to change.
- Analysis of real conversation transcripts
- Review of prompts, routing and guardrails
- Evaluations and regression tests
- Prioritised action plan
-
EU AI Act consulting for voice agents
Voice AI triggers transparency obligations and, depending on the use case, further requirements. I classify your use case and build the requirements straight into the agent.
- Role and risk classification
- AI disclosure in the conversation
- Documentation and logging
- Human oversight and escalation
Certified by ElevenLabs
In the Partner Technical Certification Program, I built a production-shaped agent that was graded by an ElevenLabs Forward Deployed Engineer.
- Certificate
- ElevenAgents Certified Partner Deployment Engineer
- Issuer
- ElevenLabs, Partner ElevenAcademy
- Issued
- 28 September 2026
- Valid until
- 28 September 2027
- Credential ID
- d8f2cd60-2cd5-493d-a96f-967101dcde39
Assessed skills
- Agent design and deployment on ElevenAgents
- Intent design
- Webhook tool integration
- Identity verification for privileged actions
- Deterministic routing on tool outcomes
- Data protection and scope guardrails
- Quality assurance and regression testing
How a project works
Small steps and results you can hear early. After each step, you decide whether to continue.
-
Intro call
We clarify goals, call volume, systems and the regulatory framework.
-
Concept
Conversation flows, tools, data flows and risk classification on one page.
-
Prototype
An agent you can call yourself within a few days.
-
Test
Evaluations with realistic cases, edge cases and abuse attempts.
-
Go-live
Gradual rollout, monitoring and handover to your team.
The EU AI Act, built into the conversation
Regulation often stays on paper. With voice agents, it has to be audible in the conversation: in the greeting, in how data is handled and in the handover to a person.
Providers of AI systems that interact directly with people must ensure those people know they are talking to an AI.
Paraphrased from Article 50(1), Regulation (EU) 2024/1689
-
Role and risk classAre you a provider or a deployer? Is the use case high-risk, for example in recruitment or credit?
-
TransparencyAn AI notice at the start of the call, clearly worded and in every language the agent speaks.
-
Human oversightClear rules for when the agent hands over to staff and how interventions are logged.
-
Documentation and AI literacyTraceable system documentation and training for the teams working with the agent.
This is technical and organisational consulting, not legal advice. If you wish, I work alongside your legal team or law firm.
Frequently asked questions
Short answers about voice agents, the certification and the EU AI Act. Missing your question? Send me an email.
What is a voice AI agent?
A voice AI agent is an AI-powered voice assistant that handles phone calls or web conversations on its own. It understands spoken language, answers in a natural voice and completes tasks through connected systems, such as booking appointments, checking orders or handing a request over to a member of staff.
What does “ElevenLabs Certified Partner Deployment Engineer” mean?
ElevenLabs awards this certification in its Partner Technical Certification Program. To earn it, I built a production-shaped agent on the ElevenAgents platform that was graded by an ElevenLabs Forward Deployed Engineer. The assessment covered tool integration, identity verification, routing, guardrails and testing. The certificate is valid until September 2027 and can be verified publicly.
Which use cases are voice agents good for?
Mainly recurring conversations with a clear goal: customer service hotlines, appointment booking, order and delivery status, self-service reports such as meter readings, lead qualification and internal help desks. When a request needs empathy or judgement, the agent hands over to a person.
Does an AI phone assistant have to say that it is an AI?
As a rule, yes. Under Article 50(1) of the EU AI Act (Regulation (EU) 2024/1689), people must be informed that they are interacting with an AI system unless this is obvious. For voice agents, the best place to do this is the greeting. Depending on the use case, further obligations apply, for example for high-risk applications.
What do you check when reviewing an existing voice agent?
I analyse real conversation transcripts and review the prompt, conversation flow, tool calls, routing and guardrails. I add evaluations with edge cases and abuse attempts. You get a prioritised action plan: what to fix now, what to address in the medium term, and what already works well.
Let’s talk about your agent.
Tell me briefly what your agent should do or where it falls short. I work with companies across the EU, in English or German, and will reply within two working days with a suggested time.
